Skip to main content
Journal of Sport and Health Science logoLink to Journal of Sport and Health Science
. 2026 Jul 23;15:101162. doi: 10.1016/j.jshs.2026.101162

Comparison of daily step counts among wearable devices in free-living adults: A systematic review and network meta-analysis

Amanda E Paluch a,b,⁎, Kelly R Evenson c, Eric M Eberl d, Susan P Marcotte e, Pardis Parvizi a, Erika Rees-Punia f, Lindsay Toth g, David R Bassett h
PMCID: PMC13744342  PMID: 42492741

Highlights

  • •

    In a network meta-analysis of 59 free-living studies, hip-worn research accelerometers, consumer wearables, and pedometers produced comparable daily step estimates, with most differences within 10% and fewer than 700 steps/day.

  • •

    Wear location was the dominant source of variation: research-grade wrist- and ankle-worn devices estimated 15%–30% more daily steps (∼1000–2200 steps/day) than hip-worn devices, while consumer wrist-worn devices showed smaller differences (3%–15%, or ∼200 to 1000 steps/day).

  • •

    Within-brand differences across device generations were generally smaller than wear location differences in daily steps, suggesting reasonable comparability across product lifecycles within the same brand and wear location.

  • •

    Algorithm selection can substantially impact daily step estimates, highlighting the need to report device brand, model, version, wear location, and processing algorithm to enable accurate interpretation and replication of findings.

Keywords: Accelerometer, Pedometer, Physical activity, Comparison studies

Abstract

Background

Wearable activity monitors are widely used to measure daily step counts, a simple and health-relevant metric of physical activity. However, differences in sensor technologies, algorithms, and wear locations across devices can result in substantial variability in step count estimates, creating challenges for comparing and harmonizing data across studies and populations. This study aimed to compare free-living daily step counts across commonly used activity monitoring devices in adults using network meta-analysis.

Methods

We systematically searched PubMed, Scopus, and PsycINFO through November 2025 for studies reporting concurrent daily step measurements using at least 2 devices from our target list of 13 manufacturers of research- and consumer-grade accelerometers and pedometers. Network meta-analysis estimated the ratio of means (ROMs) and mean differences in daily steps between devices.

Results

Across 59 studies, most consumer devices, pedometers, and the thigh (ActivPAL) and hip-worn research devices (ActiGraph GT, Axivity, and Actical) were within 10% ROM and <700 steps/day. Wear location was the primary source of variation, with wrist-worn and ankle-worn (StepWatch) research-grade accelerometers estimating 15%‒30% more daily steps (+1000 to 2200 steps/day) than hip-worn devices, while consumer-grade wrist devices showed smaller differences (3%‒15% higher or +200 to 1000 steps/day).

Conclusion

These findings provide researchers with evidence-based guidance for selecting appropriate step devices, interpreting step count data across studies, and considerations when combining data from multiple device types and wear locations.

Graphical abstract

Image, graphical abstract

1. Introduction

Wearable activity monitors have become indispensable tools for quantifying physical activity in both research and everyday life. Among the various metrics they provide, daily step count stands out for its simplicity, accessibility, and relevance to health outcomes.1 Daily steps are associated with a wide range of health outcomes including cardiovascular disease, cancer, dementia, diabetes, and mortality.2, 3, 4, 5 As the popularity of step counting surges, so has the diversity of devices available ranging from research-grade accelerometers to consumer-grade fitness trackers and smartwatches. However, this proliferation of devices has introduced new challenges in comparing and interpreting step counts across studies and populations.

A wide range of wearable devices are now commonly used to measure physical activity in research, clinical trials, and public health surveillance.6 These wearable devices differ in their underlying sensor technologies, step detection algorithms, and recommended wear locations which can result in substantial variability in step count estimates. As the use of these devices expands, understanding how their daily step outputs compare is essential for ensuring accurate interpretation of step metrics.

Previous meta-analysis research has shown even within a single brand, such as ActiGraph (Ametris, LLC, Pensacola, FL, USA) step counts can vary significantly depending on model, wear location, and filtering method.7 However, no comprehensive synthesis has compared step counts across a broad range of commonly used devices. The primary aim of this study was to compare free-living daily step counts measured among adults across a diverse set of activity monitors using a network meta-analysis. We aimed to quantify the relative differences in daily step estimates across 13 common research- and consumer-grade device manufacturers and identify patterns related to device type, model, version, algorithms, and wear location. This work provides a foundation for interpreting step-based metrics across heterogeneous studies and supports efforts to standardize step measurement in research and practice.

2. Methods

2.1. Device selection

Devices were chosen to represent the spectrum of devices used in physical activity research: research-grade accelerometers (ActiGraph including GT series and the older 7164, Actical, Axivity, GENEActiv, ActivPAL, and StepWatch), consumer-grade accelerometers commonly used in the US (Fitbit, Garmin, and Apple Watch), and pedometers (Yamax, Omron, New Lifestyles, and Accusplit). The research-grade devices are among the most commonly used accelerometers according to a recent scoping review of large observational studies.6 We included the StepWatch as it is often used in step validation research as a comparison with other devices and produces estimates most similar to direct observation.8 Approximately 1 in 3 adults in the United States owns a wearable device.9 In North America, Apple Watch devices represent the largest share of the consumer wearable market, followed by Fitbit, and Garmin.10 The Yamax, Omron, New Lifestyles, and Accusplit pedometers were included as they are among the most frequently used pedometers in physical activity research11, 12, 13, 14 and remain popular consumer options due to their affordability (<USD100) and ease of use.

2.2. Literature search

We conducted a systematic literature review following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Our objective was to identify studies reporting free-living daily step counts using at least two of the following devices: ActiGraph, Actical, Axivity, GENEActiv, ActivPAL, StepWatch, Accusplit, Omron, Yamax, Garmin, Fitbit, Apple Watch, and New Lifestyles. We searched PubMed, Scopus, and PsycINFO for articles published through November 8th 2025, using combinations of the following terms: (“ActiGraph” OR “Axivity” OR “GENEactiv” OR “ActivPAL” OR “StepWatch” OR “Actical” OR “Accusplit” OR “Omron” OR “Yamax” OR “Fitbit” OR “Garmin” OR “Apple Watch” OR “New Lifestyles”) AND (“steps” OR “step count”) AND (“valid*” OR “compar*” OR “accuracy”). The review was not registered in International Prospective Register of Systematic Reviews (PROSPERO) because there was no health outcome of interest, a requirement for registration.15

2.3. Inclusion criteria

Studies were included if they: (a) Reported daily step counts under free-living conditions; (b) Included at least 1 full day (≥10 h of wear time); (c) Included community-dwelling adults (≥18 years) without mobility impairments or disease-specific recruitment; (d) Compared at least 2 eligible devices worn concurrently; and (e) Were published in English. We excluded laboratory-only validations without free-living data, studies that did not report or allow estimation of mean and standard deviation for steps, and smartphone-only step-count evaluations.

2.4. Data extraction

Two reviewers (AEP, KRE, EME, LT, ERP, SPM, PP, or DRB) independently screened titles, abstracts, and full texts, with discrepancies resolved by a 3rd reviewer (AEP, KRE, EME, LT, ERP, SPM, PP, or DRB). From each study, we extracted study characteristics (author, year, country, sample size, and demographics), device details (brand, model, wear location, firmware, and filtering settings), protocol details (number of days, wear time, and concurrent wear), and free-living step count outcomes (mean and SD of steps/day). Authors were contacted for missing step count data. Medians and interquartile ranges were converted to means and SDs per Cochrane protocols.16 Screening and extraction used Covidence software (Veritas Health Innovation Ltd, Melbourne, VIC, Australia).17

We included only ActiGraph data using the normal filter (Low Frequency Extension (LFE), disabled), as our previous meta-analysis found LFE inflated step counts by approximately 39% on the hip.7 LFE may record non-ambulatory movements as steps, particularly during low-intensity activities, thus reducing the validity of step count estimates.

2.5. Quality assessment

We adapted a 7-item quality assessment tool developed for device validation studies.18 The tool included evaluation of clarity of the research question, population description, protocol detail, anatomical placement, device specification, data processing methods, and software reporting. Each item contributed one point to a total score (0–7) with a higher score indicating higher study quality. A detailed description of the questions is provided in Supplementary Table 1.

2.6. Statistical analysis

Given the diversity of device models, versions, algorithms, and wear locations, we grouped devices to allow more interpretable comparisons. Primary analyses grouped devices by manufacturer and wear locations—hip and wrist for most devices, ankle for StepWatch, and thigh for ActivPAL. ActiGraph devices were separated into uniaxial 7164 and triaxial GT series models based on prior evidence of measurement differences.7,19 Secondary analyses examined model (e.g., Fitbit Flex vs. Charge), version (e.g., Fitbit Flex vs. Flex 2), and algorithm (e.g., GGIR, Verisense, ActiLife) groupings, and less common wear locations (e.g., pocket).

A network meta-analysis enables a combination of direct and indirect evidence to compare multiple devices, including those never compared head-to-head.20 We conducted a network meta-analysis to estimate differences in daily step counts across wearable devices. Results were presented in 2 formats: (1) ratio of means (ROMs, Device 1/Device 2), also referred to as percent comparison, ((Device 1/Device 2) × 100%) and (2) mean difference in steps (Device 1–Device 2). For studies longer than 1 day, the average steps per day across all days was used. Estimates and 95% confidence intervals (95%CIs) were calculated using DerSimonian–Laird random-effects models.16 Heterogeneity was assessed using the I² statistic and categorized as low (<25%), moderate (25%–75%), or high (>75%).21

Forest plots used hip-worn ActiGraph GT (ActiLife algorithm) as the network reference because it was the most common comparator we found in this review, maximizing network connectivity and precision. This is a pragmatic choice, not a claim of gold-standard accuracy. Any device could serve as the reference in a network meta-analysis with identical conclusions, though with reduced precision if there were fewer direct comparisons. We generated network diagrams to visualize direct comparisons between devices. League tables with heatmap overlap were developed to quantify comparisons of step estimates across all devices.

In sensitivity analyses, we generated net split forest plots to compare direct, indirect, and overall network estimates for each treatment comparison. In this analysis, the I² statistic quantified heterogeneity across direct and indirect estimates, and the direct evidence proportion indicated the relative contribution of direct evidence to overall network estimates.22 All network meta-analysis analyses were conducted in R (Version 4.5; R Foundation for Statistical Computing, Vienna, Austria) using the netmeta package.23 For the subset of studies reporting Pearson correlation coefficients, a random-effects meta-analysis was conducted using the metafor package in R.24

3. Results

3.1. Systematic review

We identified a total of 6860 records (Fig. 1). After removing 4303 duplicates, 2557 unique studies remained for title and abstract screening. We conducted a full-text review for 104 studies. Following full-text assessment, we excluded 33 studies. A total of 71 studies met all inclusion criteria and were included in the systematic review. Of these, we excluded 12 studies19,25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35 from the meta-analysis due to insufficient data for quantitative synthesis. Therefore, the final meta-analysis included 59 studies.8,36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93

Fig. 1.

Fig 1 dummy alt text

PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow chart of free-living steps per day device comparison studies.

3.2. Quality assessment

The 71 studies8,19,25‒93 demonstrated generally high methodological quality (Supplementary Table 1). Most studies achieved a score of 7 (k = 30, 42%)8,19,25,26,30,34,36, 37, 38,40,42,45,48, 49, 50, 51,53,54,57,59,60,71,75,78,81,83,84,90,92,93 6 (k = 22, 31%)27,29,31, 32, 33,35,39,43,44,46,52,62,65,66,69,70,72,73,76,82,88,91 or 5 (k = 14, 20%)28,41,47,55,58,63,64,67,68,77,79,85,86,89 out of the maximum score of 7. Across the 7 domains, performance was strongest for clarity of research question (k = 71, 100%), protocol detail (k = 69, 97%), and wear location reporting (k = 65, 92%). Sex generalizability was adequate in most studies (k = 64, 90%).

Device specifications and age generalizability were the most frequently unmet criteria. One quarter of studies28,35,39,44,56,58,61,66,70,72,74,76,77,79,86,87,91 (k = 17, 24%) inadequately reported device specifications, including device version, software, firmware, and initialization settings. Age generalizability was the most common limitation, with 21 studies28,41,43,47,56,58,61‒65,67,68,73,74,79,80,85,86,88,89 (30%) recruiting a narrow age range, most often only college-aged adults.

3.3. Descriptive summary of studies

Studies included in the meta-analysis covered a range of sample sizes (median n = 35 (interquartile range (IQR): 21‒51); range: 9‒953) and geographic settings (17 countries). The median participant age was 33 years (IQR: 26‒52). Notably, 81% of studies (k = 48)8,36,38,39,41, 42, 43, 44, 45, 46, 47, 48, 49, 50,52,54, 55, 56, 57, 58, 59, 60, 61,64,66, 67, 68, 69, 70, 71, 72, 73, 74, 75,77, 78, 79, 80,82,83,85,87, 88, 89, 90, 91, 92, 93 included only young to middle-aged adults, while 11 studies37,40,51,53,62,63,65,72,76,81,86 included some older adults >65 years (Supplementary Table 2). Protocols ranged from 1 to 15 days of wear. Most studies (k = 39, 66%)8,38,39,45,47, 48, 49, 50,52, 53, 54, 55, 56, 57, 58, 59,64,66, 67, 68, 69, 70,73, 74, 75, 76,78, 79, 80,82, 83, 84, 85,87,88,90, 91, 92, 93 asked participants to wear the device during waking hours only, while 16 studies (27%)36,40, 41, 42, 43, 44,46,51,61, 62, 63,65,71,72,81,89 asked participants to wear devices for 24 h; and 4 studies37,60,77,86 did both with wrist devices worn for 24 h and hip for waking hours only. Supplementary Tables 2 and 3 provide additional details on study characteristics included and excluded from the network meta-analysis, respectively.

The network included 17 distinct device × wear-location nodes, 47 designs, and 169 pairwise comparisons (Fig. 2). The hip-worn ActiGraph GT was the most central node with 77 direct comparisons, followed by the Yamax hip (48 direct comparisons), StepWatch ankle (36 direct comparisons), and Fitbit wrist (34 direct comparisons). Devices with a moderate number of direct comparisons (>10 and <30) included ActiGraph 7164, ActiGraph GT wrist, ActivPAL, Fitbit hip, Garmin, New Lifestyles, and Omron. Several devices (Apple Watch, Actical, Axivity, GENEActiv, Accusplit) had 7 or fewer direct comparisons and relied on indirect evidence via common comparators (Fig. 2, Supplementary Table 4).

Fig. 2.

Fig 2 dummy alt text

Network diagram illustrating the connectivity of evidence across all device comparisons. The primary network meta-analysis included 17 treatments (e.g., device/wear locations), 47 designs, and 169 pairwise comparisons. Line thickness represents the number of direct head-to-head comparisons.

3.4. Comparative device performance

To facilitate visual interpretation and comparison, we selected the most commonly used device across studies, the hip-worn ActiGraph GT (ActiLife algorithm), as the reference standard against which all other devices were compared. Fig. 3 presents forest plots of the ROM and mean difference (MD) for each device relative to the hip-worn ActiGraph GT comparator, organized by manufacturer and wear location. Supplementary Table 4 provides a summary of the number of comparisons between devices. High heterogeneity was observed across the network (I² = 78%) (Supplementary Fig. 1).

Fig. 3.

Fig 3 dummy alt text

Forest plot of network meta-analysis comparing daily steps across devices and wear locations: (A) Ratio of means, (B) Mean differences. Comparison/Reference (bold circle): hip-worn ActiGraph GT models using the ActiLife algorithm. Larger bubbles indicate greater precision of point estimate (smaller confidence intervals).

3.4.1. Research-grade results

When compared to the hip-worn ActiGraph GT models, the hip-worn Axivity (using ActiLife algorithm) had a similar step estimate (ROM = 1.00, 95%CI: 0.73‒1.37; MD = −51 steps/day, 95%CI: −3123 to −3020), although based on indirect evidence, which is reflected in wide confidence intervals. The hip-worn uniaxial ActiGraph 7164 produced estimates 17% higher than the hip-worn ActiGraph GT (ROM = 1.17, 95%CI: 1.00‒1.37; MD = +1510 steps/day, 95%CI: 64‒2955). Wrist-worn research devices consistently produced substantially higher estimates: ActiGraph GT (wrist) yielded estimates 30% higher than the hip-worn ActiGraph GT (ROM = 1.30, 95%CI: 1.16‒1.45; MD = +2223 steps/day, 95%CI: 1284‒3161), while Axivity (wrist) and GENEActiv (wrist) showed estimates 18%‒19% higher than the hip-worn ActiGraph GT (Axivity: ROM = 1.19, 95%CI: 0.83‒1.71; GENEActiv: ROM = 1.18, 95%CI: 0.78‒1.79), though with wider confidence intervals reflecting limited direct comparisons. Alternative wear locations showed ankle-worn StepWatch produced estimates 29% higher than the hip-worn ActiGraph GT (ROM = 1.29, 95%CI: 1.17‒1.42; MD = +2123 steps/day, 95%CI: 1298‒2948) similar to wrist devices, while thigh-worn ActivPAL showed estimates 10% higher than the hip-worn ActiGraph GT (ROM = 1.10, 95%CI: 0.94‒1.29; MD = +652 steps/day, 95%CI: −657 to 1960).

3.4.2. Consumer-grade results

Consumer-grade wearable devices showed patterns similar to research-grade accelerometers, though with generally smaller wear location effects. Hip-worn Fitbit demonstrated 7% higher mean step estimates compared to the hip-worn ActiGraph GT comparator (ROM = 1.07, 95%CI: 0.92‒1.25; MD = +570 steps/day, 95%CI: −715 to 1855). Wrist-worn consumer devices produced moderately higher estimates than hip-worn devices, though the magnitude proved smaller than research accelerometers (3%‒15% higher than the hip-worn ActiGraph GT for consumer vs. 18%‒30% higher for research). Fitbit (wrist), with the strongest evidence base (16 direct comparisons), yielded estimates 15% higher than the hip-worn ActiGraph GT (ROM = 1.15, 95%CI: 1.05‒1.26; MD = +1188 steps/day, 95%CI: 363‒2013), while Garmin and Apple Watch (wrist) showed estimates 3%‒7% higher than the hip-worn ActiGraph GT (Garmin: ROM = 1.03, 95%CI: 0.90‒1.19; MD = +374 steps/day, 95%CI: −800 to 1547; Apple Watch: ROM = 1.07, 95%CI: 0.90‒1.26; MD = +483 steps/day, 95%CI: −975 to 1941).

3.4.3. Pedometer results

Three of the 4 pedometer brands demonstrated similar mean step estimates with the hip-worn ActiGraph GT comparator: Yamax showed similar estimates (ROM = 1.00, 95%CI: 0.92‒1.08; MD = −74 steps/day, 95%CI: −774 to 627), as did Omron (ROM = 1.00, 95%CI: 0.88‒1.14; MD = −55 steps/day, 95%CI: −1152 to 1042), while New Lifestyles showed estimates 6% higher than the hip-worn ActiGraph GT (ROM = 1.06, 95%CI: 0.94‒1.19; MD = +403 steps/day, 95%CI: −577 to 1383). In contrast, Accusplit produced estimates 30% lower than the hip-worn ActiGraph GT (ROM = 0.70, 95%CI: 0.51‒0.94; MD = −2321 steps/day, 95%CI: −4905 to 262). However, this finding should be interpreted with caution, as the Accusplit was not directly compared to the ActiGraph GT in any study; this estimate is based on indirect evidence, which is reflected in wide confidence intervals.

3.4.4. League table results

To provide a comprehensive summary of all possible device comparisons, Fig. 4 presents a league table heatmap showing ROM and MD estimates for every pairwise comparison in the network (see Supplementary Fig. 2 for 95%CIs).

Fig. 4.

Fig 4 dummy alt text

League table heatmap of network meta-analysis comparing daily steps across devices and wear locations: (A) Ratio of means (RoM), (B) Mean differences. Each cell compares the column device with the row device. Mean differences were calculated as the column-device estimate minus the row-device estimate, and ratios of means were calculated as the column-device estimate divided by the row-device estimate. Therefore, MD > 0 or ROM > 1 indicates that the column device records more steps than the row device, whereas MD < 0 or ROM < 1 indicates that the column device records fewer steps. Corresponding cells on opposite sides of the diagonal have opposite signs for MD and reciprocal values for ROM. Colors indicate the magnitude of difference: light blue cells show where the column device records fewer steps than the row device (ROM < 1.0 or MD < 0), white cells indicate similar step counts (ROM ≈ 1.0 or MD ≈ 0), and light red cells show where the column device records more steps than the row device (ROM > 1.0 or MD > 0).

The league table heatmap confirms the wear location gradient observed in Fig. 3. The visualization shows warmer colors (higher step estimates) for wrist- and ankle-worn devices compared to thigh- and hip-worn devices, regardless of manufacturer. Hip-worn devices clustered tightly, with most ROM values between 0.91 and 1.10, demonstrating similar group-level steps estimates across research accelerometers, consumer wearables, and pedometers. Within-brand comparisons reinforced the wear location effect for research devices, for example, ActiGraph GT wrist produced 30% more steps than ActiGraph GT hip (ROM = 1.30). There was less difference by wear location for consumer devices: for example, Fitbit wrist showed 6% higher estimates than Fitbit hip (ROM = 1.06). In general, wrist-worn research accelerometers consistently produced 15%–30% higher estimates while wrist-worn consumer devices were modestly higher (3%–15%) compared to hip-worn devices.

3.5. Sensitivity analyses results

The net split forest plot (Supplementary Fig. 3) displays ROMs for all pairwise device comparisons, with the vertical reference line indicating no difference between devices (ROM = 1.0). The number of direct comparisons ranges from 0 to 16 studies per comparison, with direct evidence proportions ranging from 0% to 100% (indicating the percentage of evidence from head-to-head studies vs. indirect network comparisons). I² values range from 0% to 87%, with most comparisons showing moderate to substantial heterogeneity (I² = 40%‒75%) between the direct and indirect estimates.

Supplementary Figs. 4‒7 present network meta-analyses of 59 device configurations and 201 pairwise comparisons with results disaggregated by device models, algorithms, versions, and alternative wear locations. The reference comparison for this analysis was the Actigraph GT3X hip-worn using the ActiLife algorithm. Within-brand differences remained consistently smaller than wear location effects. For ActiGraph GT devices, multiple generations (GT1M, GT3X, GT9X) showed ROM values of 0.99‒1.10 when worn at the hip, suggesting device upgrades have maintained relative consistency in step detection. Garmin models showed tight clustering (ROM: 1.05‒1.15 compared to ActiGraph GT3X hip), with the notable exception of the Vivosmart HR (ROM = 0.90). Apple Watch generations (Series 0, 2, and 6) demonstrated consistency across product cycles (ROM: 1.10‒1.19 compared to ActiGraph GT3X hip). Fitbit models showed some variability (ROM: 0.99‒1.27 compared to ActiGraph GT3X hip), but demonstrated consistency within specific models across versions: Charge models (ROM: 1.21‒1.26), Flex vs. Flex 2 (ROM: 0.99‒1.01), and hip-worn One and Zip (ROM: 1.09‒1.10). Yamax pedometer models showed stability across multiple Digiwalker versions (ROM: 1.02‒1.11 compared to ActiGraph GT3X hip), while the ALNESS 200S was notably lower (ROM = 0.81).

Research accelerometers using the same algorithms at the same wear location showed similar step counts both within and across brands. Notably, when using the same processing algorithms across different research-grade devices worn on the wrist, estimates were within 10% of each other. For example, the Verisense algorithms (Versions 1 and 2) were applied to data from 2 different wrist-worn devices, the ActiGraph GT and the Axivity. The resulting ROMs were similar: 1.16–1.18 for the ActiGraph GT wrist and 1.23–1.25 for the Axivity wrist (all relative to the reference of ActiGraph GT hip with ActiLife processing).

Finally, our review identified 74 reported Pearson correlations of daily steps between 2 devices, ranging from 0.45 to 0.99. Our meta-analysis showed a pooled correlation coefficient of 0.93 (95%CI: 0.91–0.94) (Supplementary Fig. 8).

4. Discussion

4.1. Main findings

This network meta-analysis of free-living daily step comparative studies of adults revealed important areas of consistency alongside key sources of variation in daily step count estimates. Within the same wear location and device type (research vs. consumer vs. pedometer), different brands and models show similar mean estimates, with most daily step comparisons less than 10% ROM and <700 steps/day. In our main analysis, all consumer devices, most pedometers, and the thigh (ActivPAL) and hip-worn research devices (ActiGraph GT, Axivity, and Actical) were within 10% of each other. However, we identified a larger systematic difference in which research wrist-worn devices consistently produce daily step estimates 15% to 30% higher (1000‒2200 steps/day) than most other devices, including hip-worn research devices, and any wear location of consumer devices and pedometers. Our results showed high heterogeneity (I² = 78%). While high heterogeneity might typically signal methodological inconsistency, in this context it is expected and interpretable; systematic differences in wear location, device type, and algorithms are precisely the sources of variation this review aimed to characterize. These patterns provide researchers with guidance for device selection and data interpretation, and considerations when combining or comparing across multiple devices.

Three devices in our primary analysis showed considerably different estimates from the broader pattern. The StepWatch ankle and the uniaxial ActiGraph 7164 appeared systematically higher and Accusplit appeared systematically lower, compared to most other devices. The StepWatch’s higher estimates (ROM: 1.20‒1.30 compared to hip-worn devices) likely reflect its ankle placement, which captures leg movements more directly than hip-worn devices. The ActiGraph 7164’s higher estimates (ROM = 1.17 compared to ActiGraph GT hip) may reflect differences in the uniaxial design and filtering algorithms compared to newer triaxial models.7 The Accusplit’s lower estimates (30%‒41% fewer steps than other hip devices) warrant cautious interpretation. Validation studies show inconsistent performance across Accusplit models, with older spring-levered versions significantly undercounting steps at slower speeds (only 26% at 1 mile per hour)79 while newer accelerometer-based models demonstrated higher accuracy (>93%).94 Our analysis included only 2 direct comparisons and could not distinguish between Accusplit models, limiting interpretation of these findings.

The magnitude and pattern of differences we observed carry important implications for device interchangeability. No established criteria currently define what constitutes sufficiently similar step counts for devices to be considered interchangeable. However, expert consensus on activity tracker validation suggests ≤5% error for clinical trials and ≤10%–15% for general population use.18 Applying this framework to our findings reveals that most hip-worn devices show similar estimates to each other, including research accelerometers (ActiGraph GT), consumer wearables (Fitbit), and most pedometers (Yamax, Omron, New Lifestyles), with differences typically within 10% (ROM: 0.90‒1.10). However, estimates diverge substantially when wear location differs: any research wrist-worn device compared to any hip-worn device showed differences ranging from 15% to 30%, exceeding even the more lenient 15% threshold. These differences equal or exceed effect sizes of many physical activity interventions.95 Wear location emerges as the most critical factor in deciding whether step counts can be directly compared across studies, with device brand playing a secondary role within the same wear location.

Several systematic reviews indicate consumer and research devices summarized the validity and reliability for physical activity measurement, primarily in lab-based settings.96, 97, 98, 99 These reviews consistently show step counts prove more accurate and reliable measures than other physical activity metrics such as minutes of moderate-to-vigorous physical activity or energy expenditure. 96, 97, 98, 99 However, validation protocols and criterion comparisons vary substantially across studies, precluding conclusions about whether validated step counters produce similar estimates. The current analysis addresses a different translational research question: how do devices directly compare to each other in real-world applications? This approach provides actionable information for researchers selecting devices, interpreting existing literature, and assessing the interchangeability of step counting devices in clinical and public health settings.

To our knowledge, this represents the first network meta-analysis of daily step device estimates, substantially expanding the comparative evidence base. We quantified wear location effects systematically across multiple device types, demonstrating wear location explains more variance than device brand, model and version. We examined model and algorithm specificity systematically, showing within-brand differences remain minimal compared to wear location effects. We included a variety of the most current research- and consumer-grade devices, addressing a critical gap given their increasing use in research and by the public.

4.2. What is a step?

Step counting methods vary substantially in how they detect and quantify steps. Most current step counters use micro-electrical mechanical system (MEMS) accelerometers to sense motion.100 These devices record acceleration and/or angle of inclination subsequently filtering the raw signal(s) and applying algorithms that use peak detection methods to count steps. Pedometers count steps by detecting vertical acceleration during gait, using either a horizontal, spring-suspended lever arm or an electronic accelerometer-based sensor.99

Devices count steps in a relatively straightforward manner during continuous walking, as walking represents a harmonic motion with steps occurring within a predictable frequency range (approximately 1–2 Hz).101 Wearable devices often employ filters to attenuate frequencies outside this range, reducing acceleration peaks from extraneous sources such as vibrations from operating a motor vehicle. Some devices require several successive peaks at regular intervals before classifying the detected peaks as steps or have a minimum threshold for the time interval between steps.99 This may systematically undercount steps during gait initiation, termination, and other non-steady-state walking. Counting steps becomes more challenging during intermittent stepping activities.102 During household chores (e.g., vacuuming, meal preparation) and certain recreational activities (e.g., tennis), steps occur at irregular intervals and often involve multi-directional movements. These characteristics make step detection more variable and error prone.44 Consequently, intermittent activities show substantially more variability in step count accuracy compared to continuous walking.103

The precise methods devices use for step detection may vary by wear location. The ankle-worn StepWatch uses forward acceleration or heel lift to detect steps, enabling detection of slow, shuffling steps which other devices might miss.104,105 The thigh-worn ActivPAL uses the thigh angle of inclination to detect steps, counting each transition from positive to negative angle as the thigh moves through its walking cycle.106 Most step count algorithms are proprietary; however, some research-grade accelerometers record raw acceleration and utilize open-source algorithms, allowing greater transparency and reproducibility.107, 108, 109, 110, 111, 112 The advantage of open-source algorithms becomes evident when comparing research devices worn at the same location using identical processing methods: in this network meta-analysis, the ActiGraph GT and Axivity devices worn on the wrist showed similar step estimates (within 2%‒6%) when processed with the same Verisense algorithm, demonstrating step data may be harmonized across devices when wear locations are consistent and standardized algorithms are applied.

4.3. Wear location

The systematic pattern of higher step estimates from wrist-worn compared to hip-worn devices may reflect detection of non-ambulatory arm movements as steps.100 During many daily activities, including gestures while talking, food preparation, and upper-body exercise, the wrist undergoes some accelerations meeting step detection thresholds despite the absence of ambulatory activity.113 Hip-worn devices, positioned closer to the body’s center of mass, experience accelerations predominantly during actual locomotion and remain less influenced by isolated limb movements.71 However, it is also plausible wrist- and ankle-worn devices capture true ambulatory activity. The StepWatch, which most closely matches direct observation,8 produces estimates similar to wrist-worn research accelerometers, suggesting higher estimates from these devices may reflect genuine step detection rather than overcounting. Whether these differences reflect genuine capture of additional ambulatory activity or overcounting of non-ambulatory movements remains unresolved.

The league table heatmap (Fig. 4) provided visual confirmation of the influence of wear location through several patterns. First, the consistency of color patterns across device types: when comparing wrist-worn devices (columns) to hip-worn devices (rows), the upper triangle shows ROM values consistently >1.0 (warm colors), indicating wrist-worn devices record more steps regardless of manufacturer. Second, the ActiGraph GT device demonstrates this within-brand: comparing ActiGraph GT wrist (column) to ActiGraph GT hip (row) yields ROM = 1.30, meaning the wrist device records 30% more steps. Third, the ankle worn StepWatch records 20%‒30% more steps than hip devices. However, the thigh-worn ActivPAL yielded ROMs within ±10% compared to most other devices indicating intermediate estimates. These findings carry substantial practical implications. We do not recommend direct comparison of absolute step counts across studies using different wear locations without calibration. For optimal within-study consistency, intervention and longitudinal observational studies should use identical devices and wear locations at all time points.

4.4. Device type: research vs. consumer vs. pedometer

Within each wear location category, device types showed distinct patterns. Among hip-worn devices, research accelerometers, consumer wearables, and most pedometers demonstrated similar mean estimates (ROM: 0.91‒1.10), with ActiGraph GT hip, Axivity hip, Fitbit hip, Yamax, Omron, and New Lifestyles all clustering tightly together. The notable exception, Accusplit, produced substantially lower estimates (ROM: 0.59‒0.70 when compared to other hip-worn devices), though this finding requires cautious interpretation given the device relied entirely on indirect evidence.

Among wrist-worn devices, research-grade accelerometers produced higher step estimates (ROM: 1.18–1.30 compared to the hip-worn ActiGraph GT) than consumer wearables (ROM: 1.03–1.15 compared to the hip-worn ActiGraph GT). Notably, wrist-worn consumer devices produced estimates more similar to hip-worn research and consumer accelerometers and pedometers than did wrist-worn research accelerometers. This pattern may reflect differences in algorithm design philosophy. Consumer devices may incorporate proprietary algorithms designed to filter non-ambulatory movements, providing users with step estimates more consistent with their expectations or traditional pedometer values. Research accelerometers, in contrast, often prioritize sensitivity to all movements meeting specified acceleration thresholds, accepting higher step counts in exchange for capturing more movement. Neither approach proves inherently correct; rather, they represent different tradeoffs between sensitivity and specificity for detecting ambulatory movement.

4.5. Model and algorithm specificity

The analysis incorporating detailed device configurations provided important insights across model-, version-, and algorithm-level step estimates. Models and successive device generations within brands generally demonstrated similar step estimates relative to the ActiGraph GT3X hip reference, suggesting reasonable generalizability across product lifecycles. Within-brand differences were consistently smaller than the effects of wear location, underscoring that placement may contribute more to variability than incremental hardware or firmware upgrades. Although some variability was observed across different models within a manufacturer (e.g., certain Fitbit models), successive versions within the same product line tended to maintain measurement consistency.

Additionally, the same algorithms when applied to different research devices at the same wear location produce similar estimates. This is shown with similar estimates from Verisense algorithms using ActiGraph and Axivity and the wrist. However, different algorithms had a substantial impact on daily step estimates. The Axivity AX3 accelerometer illustrates this: when worn on the hip, processing the same raw data with ActiLife software yielded a ROM of 1.07, while GGIR software yielded a ROM of 0.72, both compared to reference ActiGraph GT hip ActiLife estimates. A similar pattern was observed with the ActiGraph GT3X at the wrist, where the ActiLife algorithm produced a ROM of 1.27 compared to 1.16–1.18 from Verisense algorithms, compared to the reference ActiGraph GT hip ActiLife estimates. Our meta-analysis focused on between-device differences; other studies provide deeper insight into within-device algorithm differences. An analysis of NHANES 2011‒2014 wrist accelerometry data applying five different step counting algorithms to identical raw data found mean daily step counts varied dramatically across algorithms (range: 2453‒12,169 steps/day), despite high correlations (median r = 0.91) between methods.111 These findings underscore that algorithm selection represents an independent methodological decision with impact equal to or exceeding device selection, and should be reported with the same rigor as device specifications.

4.6. Comparator device selection

We selected the ActiGraph GT hip-worn accelerometer as the network comparator for pragmatic reasons related to this network meta-analysis rather than assumptions about superior validity. This device served as the most frequently used comparison device in validation studies, which maximized network connectivity and minimized reliance on indirect pathways. Additionally, numerous large cohort studies and intervention trials use ActiGraph devices, making it a useful reference point for researchers comparing findings across studies.6

The hip-worn ActiGraph GT is not intended in this analysis to represent a gold standard or criterion measure for step counting. Accelerometers, including the ActiGraph, fundamentally detect accelerations, not steps. The step counts represent derived estimates depending on algorithm decision rules.18 The observed differences reflect variation in what devices measure and how data are processed, not measurement error relative to true values. A device producing higher or lower estimates than the hip-worn ActiGraph GT does not necessarily lack accuracy; rather, it may detect different movement patterns or apply different classification rules. No scalable criterion measure for free-living step counting currently exists; direct observation, even with video recording, is challenging for full-day monitoring and unavailable at the scale needed to anchor a network of real-world comparisons.

4.7. Implications for research and practice

Understanding systematic device measurement similarities and differences becomes essential for interpretation of step data, comparison of findings across studies, and application of step-based guidelines and surveillance in research and public health practice.

For individual-level monitoring, consistency in device use over time matters more than absolute values for tracking personal progress. An increase in steps using any single device indicates improvement in physical activity levels. Encouragingly, our results show consistency within manufacturer models across versions, allowing comparisons across device generations within the same brand and wear location.

For within-study consistency, researchers should use identical devices and wear locations at all measurement timepoints. Wear location effects (1000–2000 daily steps) can equal or exceed effect sizes of physical activity interventions.95 If device changes become unavoidable, calibration techniques should be considered. Researchers should report device brand, model, version, algorithm, and wear location to enable replication and cross-study comparison.

When comparing across studies using different devices or wear locations, caution with absolute step counts is warranted. However, the high correlation between devices (r = 0.93) suggests relative comparisons are acceptable. Researchers should consider presenting results as percent change or relative to comparison groups rather than absolute values. Harmonization methods using study-level quantiles can facilitate meta-analysis across measurement approaches.5 However, the high correlation between devices only supports relative comparisons across studies, though correlation alone does not imply individual-level interchangeability as devices may track change similarly while differing substantially at the individual level. In the future, collection of new individual-level data on a more diverse subject pool would provide further calibration opportunities for harmonization. The current network meta-analysis findings may enable after-the-fact calibration, though this approach requires validation.

To illustrate how device-related differences may impact interpretation of population-level step counts, we present examples from 4 large cohort studies, each including both younger and older adults. U.S. NHANES 2005‒2006 using hip-worn ActiGraph 7164 reported a mean of 9676 steps/day,114 while the U.S. All of Us Study using Fitbit devices reported a median of 7731 steps/day.115 The UK Biobank using wrist-worn Axivity accelerometers reported a median of 9156 steps/day,107 and the Norwegian National Physical Activity Surveillance study using waist-worn ActiGraph GT1M reported a mean of 8002 steps/day.116 While population characteristics contribute to this variation, the range is similar in magnitude to the wear location differences we observed (1000–2000 steps/day for wrist vs. hip), suggesting device-related factors contribute meaningfully alongside true population differences in explaining variation across studies.

Standardized validation approaches would address many challenges our analysis revealed regarding heterogeneous methods and cross-device comparability. While the Consumer Technology Association established pre-production guidelines for manufacturers,117 these relied on treadmill protocols and not free-living activity. Recent efforts have proposed comprehensive standardized protocols addressing target populations, criterion measures, validation conditions, data processing, and statistical analysis,18,118 aiming to ensure methodological consistency and facilitate cross-study device comparisons.

4.8. Limitations

Several limitations warrant consideration. This network meta-analysis quantifies relative differences between devices rather than absolute accuracy, as no scalable criterion measure for free-living daily step counting currently is available. Since the review focused on device metric comparability and not health outcomes, prospective registration, such as PROSPERO, was not completed. Some devices had limited direct comparison evidence, requiring reliance on indirect estimates through the network structure, which may be less reliable than head-to-head comparisons.119 Several widely used research devices (GENEActiv, Axivity) and newer consumer devices (Apple Watch) have limited comparison data despite widespread use.107,120 The supplemental analysis with more detailed device comparisons has a notable limitation: as devices were stratified into more specific configurations, fewer direct comparisons were available per device. The evidence base comes predominantly from healthy younger adults in high-income countries with normal gait; generalizability to older adults, clinical populations, those with mobility limitations, or low- and middle-income populations remains uncertain.96,121 We did not address context specificity. Device estimates may vary by activity type, walking speed, or terrain.122,123 Our review excluded studies comparing algorithms within a single device. Nearly a quarter of studies (24%) did not adequately report data processing details such as firmware, initialization or algorithm settings; additionally, the exact anatomical placement, including wrist dominance and body side was not consistently reported, precluding more granular sensitivity analyses. Protocols varied in waking-hours vs. 24-h wear requirements; overnight wear is unlikely to meaningfully inflate step counts, though minor variability may arise from steps taken around device placement and removal. We used only study-level means and standard deviations, limiting other comparison metrics (e.g., mean absolute percent error, limits of agreement).124 Some of the devices had few direct comparisons, resulting in wide confidence intervals and requiring cautious interpretation.

4.9. Research priorities

This analysis reveals several critical research priorities. First, there is a need to expand direct device comparisons for understudied yet commonly used devices, such as the GENEActiv, Axivity, and Apple Watch, as well as other device categories, including insoles, smart rings, and smartphones. This is particularly essential given the widespread consumer use of many of these devices (e.g., the Apple Watch) and their role in some of the largest accelerometer-based research studies (e.g., the Axivity in the UK Biobank). Second, researchers need population-specific validation in older adults and individuals with mobility impairments or chronic diseases affecting gait. Geographic and demographic diversity is also necessary, as typical walking speeds and activity types, context, and patterns vary across regions and populations.125 Third, development of calibration equations to convert step counts between devices or wear locations would facilitate cross-study comparisons. Fourth, standardized validation and reporting protocols of device specifications and analytic results would improve transparency and comparability of step-related research.

5. Conclusion

Daily step counts increasingly inform clinical decision-making and public health policy, yet questions remain about comparability across the diverse array of available monitoring devices. This network meta-analysis revealed reassuring consistency within similar measurement approaches: hip-worn research accelerometers, consumer devices, and most pedometers showed comparable mean estimates, and within-brand differences across device generations remained minimal (<10% or <700 steps/day). However, wear location emerged as a critical source of systematic variation, with research-grade wrist-worn and ankle-worn devices consistently producing ∼15%–30% higher estimates and consumer wrist-worn devices producing 3%–15% higher estimates than hip-worn devices. As the field continues to evolve with new devices and algorithms emerging regularly, the research community must prioritize calibration approaches, work toward greater standardization, and acknowledge current devices measure related but not identical constructs. Ongoing validation research, calibration methods, and reporting standards will prove essential to ensure measurement innovation translates to improved physical activity surveillance, intervention research, and ultimately, public health impact.

Authors’ contributions

AEP conceived the study, participated in the systematic literature review, performed the statistical analysis, interpreted the data and drafted the manuscript; EME, SPM, ERP, and LT participated in the systematic literature review, interpreted the data, and helped draft the manuscript; PP helped to draft the manuscript; KRE and DRB conceived the study, participated in the systematic literature review, interpreted the data, and helped draft the manuscript. All authors have read and approved the final version of the manuscript, and agree with the order of presentation of the authors.

Declaration of competing interest

The authors declare that they have no competing interests.

Acknowledgments

Data sharing statement

Template data collection forms, extracted study data, analytic datasets, and code are available upon reasonable request to the corresponding author.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work the authors used Claude (Anthropic) in order to improve clarity, reduce redundancy, and check for grammatical errors in the manuscript. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.

Footnotes

Peer review under responsibility of Shanghai University of Sport.

Supplementary materials associated with this article can be found in the online version at doi:10.1016/j.jshs.2026.101162.

Supplementary materials

mmc1.pdf (6.1MB, pdf)
Download video file (6.1MB, mp4)

References

  • 1.Paluch A.E., Matthews C.E., Doherty A., et al. Daily steps as a public health metric for physical activity monitoring and promotion. Exerc Sport Sci Rev. 2026;54:15–25. doi: 10.1249/JES.0000000000000375. [DOI] [PubMed] [Google Scholar]
  • 2.Paluch A.E., Bajpai S., Ballin M., et al. Prospective association of daily steps with cardiovascular disease: A harmonized meta-analysis. Circulation. 2023;147:122–131. doi: 10.1161/CIRCULATIONAHA.122.061288. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Shreves A.H., Small S.R., Walmsley R., et al. Amount and intensity of daily total physical activity, step count, and risk of incident cancer in the UK Biobank. Br J Sports Med. 2025;59:839–847. doi: 10.1136/bjsports-2024-109360. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Ding D., Nguyen B., Nau T., et al. Daily steps and health outcomes in adults: A systematic review and dose-response meta-analysis. Lancet Public Health. 2025;10:e668–e681. doi: 10.1016/S2468-2667(25)00164-1. [DOI] [PubMed] [Google Scholar]
  • 5.Paluch A.E., Bajpai S., Bassett D.R., et al. Daily steps and all-cause mortality: A meta-analysis of 15 international cohorts. Lancet Public Health. 2022;7:e219–e228. doi: 10.1016/S2468-2667(21)00302-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Evenson K.R., Scherer E., Peter K.M., Cuthbertson C.C., Eckman S. Historical development of accelerometry measures and methods for physical activity and sedentary behavior research worldwide: A scoping review of observational studies of adults. PLoS One. 2022;17 doi: 10.1371/journal.pone.0276890. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Toth L.P., Paluch A.E., Bassett D.R., Jr, et al. Comparative analysis of ActiGraph step counting methods in adults: A systematic literature review and meta-analysis. Med Sci Sports Exerc. 2024;56:53–62. doi: 10.1249/MSS.0000000000003282. [DOI] [PubMed] [Google Scholar]
  • 8.Toth L.P., Park S., Springer C.M., Feyerabend M.D., Steeves J.A., Bassett D.R. Video-recorded validation of wearable step counters under free-living conditions. Med Sci Sports Exerc. 2018;50:1315–1322. doi: 10.1249/MSS.0000000000001569. [DOI] [PubMed] [Google Scholar]
  • 9.Dhingra L.S., Aminorroaya A., Oikonomou E.K., et al. Use of wearable devices in individuals with or at risk for cardiovascular disease in the US, 2019 to 2020. JAMA Netw Open. 2023;6 doi: 10.1001/jamanetworkopen.2023.16634. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Mordor Intelligence. North America smartwatch market‒share and brands. Accessed 27 January, 2026. https://www.mordorintelligence.com/industry-reports/north-america-smart-watch-market.
  • 11.Kraus W.E., Yates T., Tuomilehto J., et al. Relationship between baseline physical activity assessed by pedometer count and new-onset diabetes in the NAVIGATOR trial. BMJ Open Diabetes Res Care. 2018;6 doi: 10.1136/bmjdrc-2018-000523. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Dwyer T., Pezic A., Sun C., et al. Objectively measured daily steps and subsequent long term all-cause mortality: The Tasped prospective cohort study. PLoS One. 2015;10 doi: 10.1371/journal.pone.0141274. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Storti K.L., Arena V.C., Barmada M.M., et al. Physical activity levels in American-Indian adults: The strong heart family study. Am J Prev Med. 2009;37:481–487. doi: 10.1016/j.amepre.2009.07.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Kikuchi H., Fukushima N., Amagasa S., Inoue S. Temporal trends and projected daily step count from 213,924 pedometer data in a nationally representative Japanese population. Med Sci Sports Exerc. 2025;57:1763–1768. doi: 10.1249/MSS.0000000000003693. [DOI] [PubMed] [Google Scholar]
  • 15.Booth A., Clarke M., Dooley G., et al. The nuts and bolts of PROSPERO: An international prospective register of systematic reviews. Syst Rev. 2012;1:2. doi: 10.1186/2046-4053-1-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Higgins J.P.T., Li T., Deeks J.J., et al. In: Cochrane Handbook for systematic reviews of interventions. Higgins J.P.T., Thomas J., Chandler J., et al., editors. Cochrane; Oxford: 2026. Chapter 6: Choosing effect measures and computing estimates of effect. editors. [Google Scholar]
  • 17.Veritas Health Innovation. Covidence systematic review software, Melbourne, Australia: Veritas Health Innovation; 2020. Accessed 3 February, 2026. https://www.covidence.org.
  • 18.Johnston W., Judice P.B., García P.M., et al. Recommendations for determining the validity of consumer wearable and smartphone step count: Expert statement and checklist of the INTERLIVE network. Br J Sports Med. 2021;55:780–793. doi: 10.1136/bjsports-2020-103147. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Cain K.L., Conway T.L., Adams M.A., Husak L.E., Sallis J.F. Comparison of older and newer generations of ActiGraph accelerometers with the normal filter and the low frequency extension. Int J Behav Nutr Phys Act. 2013;10:51. doi: 10.1186/1479-5868-10-51. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Rouse B., Chaimani A., Li T. Network meta-analysis: An introduction for clinicians. Intern Emerg Med. 2017;12:103–111. doi: 10.1007/s11739-016-1583-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Higgins J.P., Thompson S.G., Deeks J.J., Altman D.G. Measuring inconsistency in meta-analyses. BMJ. 2003;327:557–560. doi: 10.1136/bmj.327.7414.557. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Higgins J.P.T., Caldwell D.M., Li T., Page M.J., Salanti G., et al. In: Cochrane handbook for systematic reviews of interventions. Higgins J.P.T., Thomas J., Chandler J., et al., editors. Cochrane; Oxford: 2026. Chapter 11: Undertaking network meta-analyses. editors. [Google Scholar]
  • 23.Rücker G, Krahn U, König J, et al. Netmeta: Network meta-analysis using frequentist methods, R package. Accessed 26 January, 2026. https://cran.r-project.org/web/packages/netmeta/.
  • 24.Viechtbauer W. Conducting meta-analyses in R with the metafor package. J Stat Softw. 2010;36:1–48. doi: 10.18637/jss.v036.i03. [DOI] [Google Scholar]
  • 25.Silcott N.A., Bassett D.R., Jr, Thompson D.L., Fitzhugh E.C., Steeves J.A. Evaluation of the Omron HJ-720ITC pedometer under free-living conditions. Med Sci Sports Exerc. 2011;43:1791–1797. doi: 10.1249/MSS.0b013e318212888c. [DOI] [PubMed] [Google Scholar]
  • 26.Rosenberger M.E., Buman M.P., Haskell W.L., McConnell M.V., Carstensen L.L. Twenty-four hours of sleep, sedentary behavior, and physical activity with nine wearable devices. Med Sci Sports Exerc. 2016;48:457–465. doi: 10.1249/MSS.0000000000000778. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Schneider P.L., Crouter S., Bassett D.R. Pedometer measures of free-living physical activity: Comparison of 13 models. Med Sci Sports Exerc. 2004;36:331–335. doi: 10.1249/01.MSS.0000113486.60548.E9. [DOI] [PubMed] [Google Scholar]
  • 28.Paul S.S., Tiedemann A., Hassett L.M., et al. Validity of the Fitbit activity tracker for measuring steps in community-dwelling older adults. BMJ Open Sport Exerc Med. 2015;1 doi: 10.1136/bmjsem-2015-000013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Gomersall S.R., Ng N., Burton N.W., Pavey T.G., Gilson N.D., Brown W.J. Estimating physical activity and sedentary behavior in a free-living context: A pragmatic comparison of consumer-based activity trackers and ActiGraph accelerometry. J Med Internet Res. 2016;18 doi: 10.2196/jmir.5531. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Ferguson T., Rowlands A.V., Olds T., Maher C. The validity of consumer-level, activity monitors in healthy adults worn in free-living conditions: A cross-sectional study. Int J Behav Nutr Phys Act. 2015;12:42. doi: 10.1186/s12966-015-0201-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Wanner M., Martin B.W., Meier F., Probst-Hensch N., Kriemler S. Effects of filter choice in GT3X accelerometer assessments of free-living activity. Med Sci Sports Exerc. 2013;45:170–177. doi: 10.1249/MSS.0b013e31826c2cf1. [DOI] [PubMed] [Google Scholar]
  • 32.Dondzila C.J., Swartz A.M., Miller N.E., Lenz E.K., Strath S.J. Accuracy of uploadable pedometers in laboratory, overground, and free-living conditions in young and older adults. Int J Behav Nutr Phys Act. 2012;9 doi: 10.1186/1479-5868-9-143. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Feito Y., Garner H.R., Bassett D.R. Evaluation of ActiGraph’s low-frequency filter in laboratory and free-living environments. Med Sci Sports Exerc. 2015;47:211–217. doi: 10.1249/MSS.0000000000000395. [DOI] [PubMed] [Google Scholar]
  • 34.Brewer W., Swanson B.T., Ortiz A. Validity of Fitbit’s active minutes as compared with a research-grade accelerometer and self-reported measures. BMJ Open Sport Exerc Med. 2017;3 doi: 10.1136/bmjsem-2017-000254. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.An H.S., Jones G.C., Kang S.K., Welk G.J., Lee J.M. How valid are wearable physical activity trackers for measuring steps? Eur J Sport Sci. 2017;17:360–368. doi: 10.1080/17461391.2016.1255261. [DOI] [PubMed] [Google Scholar]
  • 36.Bai Y., Tompkins C., Gell N., Dione D., Zhang T., Byun W. Comprehensive comparison of Apple Watch and Fitbit monitors in a free-living setting. PLoS One. 2021;16 doi: 10.1371/journal.pone.0251975. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Barreira T.V., Brouillette R.M., Foil H.C., Keller J.N., Tudor-Locke C. Comparison of older adults’ steps per day using NL-1000 pedometer and two GT3X+ accelerometer filters. J Aging Phys Act. 2013;21:402–416. doi: 10.1123/japa.21.4.402. [DOI] [PubMed] [Google Scholar]
  • 38.Barriera T.V., Tudor-Locke C., Champagne C.M., Broyles S.T., Johnson W.D., Katzmarzyk P.T. Comparison of GT3X accelerometer and YAMAX pedometer steps/day in a free-living sample of overweight and obese adults. J Phys Act Health. 2013;10:263–270. doi: 10.1123/jpah.10.2.263. [DOI] [PubMed] [Google Scholar]
  • 39.Breteler M.J., Janssen J.H., Spiering W., Kalkman C.J., van Solinge W.W., Dohmen D.A. Measuring free-living physical activity with three commercially available activity monitors for telemonitoring purposes: Validation study. JMIR Form Res. 2019;3 doi: 10.2196/11489. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Briggs B.C., Hall K.S., Jain C., Macrea M., Morey M.C., Oursler K.K. Assessing moderate to vigorous physical activity in older adults: Validity of a commercial activity tracker. Front Sports Act Living. 2021;3 doi: 10.3389/fspor.2021.766317. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Buchan D.S. Comparison of stepping-based metrics from ActiGraph accelerometers worn concurrently on the non-dominant wrist and waist among young adults. J Sports Sci. 2024;42:1664–1672. doi: 10.1080/02640414.2024.2404784. [DOI] [PubMed] [Google Scholar]
  • 42.Chu A.H., Ng S.H., Paknezhad M., et al. Comparison of wrist-worn Fitbit Flex and waist-worn ActiGraph for measuring steps in free-living adults. PLoS One. 2017;12 doi: 10.1371/journal.pone.0172535. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Clevenger K.A., Molesky M.J., Vusich J., Montoye A.H.K. Free-living comparison of physical activity and sleep data from Fitbit activity trackers worn on the dominant and nondominant wrists. Meas Phys Educ Exerc Sci. 2019;23:194–204. doi: 10.1080/1091367X.2019.1577737. [DOI] [Google Scholar]
  • 44.Coffman M.J., Reeve C.L., Butler S., Keeling M., Talbot L.A. Accuracy of the Yamax CW-701 pedometer for measuring steps in controlled and free-living conditions. Digit Health. 2016;2 doi: 10.1177/2055207616652526. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Crouter S.E., Schneider P.L., Karabulut M., Bassett D.R. Validity of 10 electronic pedometers for measuring steps, distance, and energy cost. Med Sci Sports Exerc. 2003;35:1455–1460. doi: 10.1249/01.MSS.0000078932.61440.A2. [DOI] [PubMed] [Google Scholar]
  • 46.Crowley P., Skotte J., Stamatakis E., et al. Comparison of physical behavior estimates from three different thigh-worn accelerometers brands: A proof-of-concept for the Prospective Physical Activity, Sitting, and Sleep consortium (ProPASS) Int J Behav Nutr Phys Act. 2019;16:65. doi: 10.1186/s12966-019-0835-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Cuberek R., El Ansari W., Fromel K., Skalik K., Sigmund E. A comparison of two motion sensors for the assessment of free-living physical activity of adolescents. Int J Environ Res Public Health. 2010;7:1558–1576. doi: 10.3390/ijerph7041558. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.De Cocker K.A., De Meyer J., De Bourdeaudhuij I.M., Cardon G.M. Non-traditional wearing positions of pedometers: Validity and reliability of the Omron HJ-203-ED pedometer under controlled and free-living conditions. J Sci Med Sport. 2012;15:418–424. doi: 10.1016/j.jsams.2012.02.002. [DOI] [PubMed] [Google Scholar]
  • 49.Degroote L., De Bourdeaudhuij I., Verloigne M., Poppe L., Crombez G. The accuracy of smart devices for measuring physical activity in daily life: Validation study. JMIR Mhealth Uhealth. 2018;6 doi: 10.2196/10972. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Degroote L., Hamerlinck G., Poels K., et al. Low-cost consumer-based trackers to measure physical activity and sleep duration among adults in free-living conditions: Validation study. JMIR Mhealth Uhealth. 2020;8 doi: 10.2196/16674. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Delobelle J., Lebuf E., Dyck D.V., et al. Fitbit’s accuracy to measure short bouts of stepping and sedentary behaviour: Validation, sensitivity and specificity study. Digit Health. 2024;10 doi: 10.1177/20552076241262710. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Dominick G.M., Winfree K.N., Pohlig R.T., Papas M.A. Physical activity assessment between consumer- and research-grade accelerometers: A comparative study in free-living conditions. JMIR Mhealth Uhealth. 2016;4 doi: 10.2196/mhealth.6281. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Farina N., Lowry R.G. The validity of consumer-level activity monitors in healthy older adults in free-living conditions. J Aging Phys Act. 2018;26:128–135. doi: 10.1123/japa.2016-0344. [DOI] [PubMed] [Google Scholar]
  • 54.Feito Y., Bassett D.R., Thompson D.L. Evaluation of activity monitors in controlled and free-living environments. Med Sci Sports Exerc. 2012;44:733–741. doi: 10.1249/MSS.0b013e3182351913. [DOI] [PubMed] [Google Scholar]
  • 55.Feito Y., Hornbuckle L.M., Reid L.A., Crouter S.E. Effect of ActiGraph’s low frequency extension for estimating steps and physical activity intensity. PLoS One. 2017;12 doi: 10.1371/journal.pone.0188242. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Garza J.L., Wu Z.H., Singh M., Cherniack M.G. Comparison of the wrist-worn Fitbit Charge 2 and the waist-worn ActiGraph GTX3 for measuring steps taken in occupational settings. Ann Work Expo Health. 2022;66:281–284. doi: 10.1093/annweh/wxab065. [DOI] [PubMed] [Google Scholar]
  • 57.Hargens T.A., Deyarmin K.N., Snyder K.M., Mihalik A.G., Sharpe L.E. Comparison of wrist-worn and hip-worn activity monitors under free living conditions. J Med Eng Technol. 2017;41:200–207. doi: 10.1080/03091902.2016.1271046. [DOI] [PubMed] [Google Scholar]
  • 58.Hickey A., John D., Sasaki J.E., Mavilia M., Freedson P. Validity of activity monitor step detection is related to movement patterns. J Phys Act Health. 2016;13:145–153. doi: 10.1123/jpah.2015-0203. [DOI] [PubMed] [Google Scholar]
  • 59.Höchsmann C., Knaier R., Infanger D., Schmidt-Trucksäss A. Validity of smartphones and activity trackers to measure steps in a free-living setting over three consecutive days. Physiol Meas. 2020;41 doi: 10.1088/1361-6579/ab635f. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Holm I., Fridolfsson J., Börjesson M., Arvidsson D. Fourteen days free-living evaluation of an open-source algorithm for counting steps in healthy adults with a large variation in physical activity level. BMC Biomed Eng. 2023;5:3. doi: 10.1186/s42490-023-00071-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61.Hong K.R., Hwang I.W., Kim H.J., Yang S.H., Lee J.M. Apple Watch 6 vs. Galaxy Watch 4: A validity study of step-count estimation in daily activities. Sensors (Basel) 2024;24 doi: 10.3390/s24144658. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62.Hyde E.T., Nguyen S., Tuz-Zahra F., et al. Agreement of step-based metrics from ActiGraph and ActivPAL accelerometers worn concurrently among older adults. J Meas Phys Behav. 2022;5:242–251. doi: 10.1123/jmpb.2022-0001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 63.Kamada M., Shiroma E.J., Harris T.B., Lee I.M. Comparison of physical activity assessed using hip- and wrist-worn accelerometers. Gait Posture. 2016;44:23–28. doi: 10.1016/j.gaitpost.2015.11.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64.Karabulut M., Crouter S.E., Bassett D.R., Jr Comparison of two waist-mounted and two ankle-mounted electronic pedometers. Eur J Appl Physiol. 2005;95:335–343. doi: 10.1007/s00421-005-0018-3. [DOI] [PubMed] [Google Scholar]
  • 65.Kastelic K., Dobnik M., Löfler S., Hofer C., Šarabon N. Validity, reliability and sensitivity to change of three consumer-grade activity trackers in controlled and free-living conditions among older adults. Sensors (Basel) 2021;21 doi: 10.3390/s21186245. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 66.Le Masurier G.C., Lee S.M., Tudor-Locke C. Motion sensor accuracy under controlled and free-living conditions. Med Sci Sports Exerc. 2004;36:905–910. doi: 10.1249/01.mss.0000126777.50188.73. [DOI] [PubMed] [Google Scholar]
  • 67.Lee K.Y., Macfarlane D.J., Cerin E. Comparison of three models of ActiGraph accelerometers during free living and controlled laboratory conditions. Eur J Sport Sci. 2013;13:332–339. doi: 10.1080/17461391.2011.643925. [DOI] [PubMed] [Google Scholar]
  • 68.Lee J.A., Williams S.M., Brown D.D., Laurson K.R. Concurrent validation of the ActiGraph GT3X+, Polar Active accelerometer, Omron HJ-720 and Yamax Digiwalker SW-701 pedometer step counts in lab-based and free-living settings. J Sports Sci. 2015;33:991–1000. doi: 10.1080/02640414.2014.981848. [DOI] [PubMed] [Google Scholar]
  • 69.Liew S.J., Gorny A.W., Tan C.S., Müller-Riemenschneider F. A mobile health team challenge to promote stepping and stair climbing activities: Exploratory feasibility study. JMIR Mhealth Uhealth. 2020;8 doi: 10.2196/12665. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 70.Lipert A., Jegier A. Comparison of different physical activity measurement methods in adults aged 45 to 64 years under free-living conditions. Clin J Sport Med. 2017;27:400–408. doi: 10.1097/JSM.0000000000000362. [DOI] [PubMed] [Google Scholar]
  • 71.Mandigout S., Lacroix J., Perrochon A., Svoboda Z., Aubourg T., Vuillerme N. Comparison of step count assessed using wrist- and hip-worn ActiGraph GT3X in free-living conditions in young and older adults. Front Med (Lausanne) 2019;6 doi: 10.3389/fmed.2019.00252. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 72.Maylor B.D., Edwardson C.L., Dempsey P.C., et al. Stepping towards more intuitive physical activity metrics with wrist-worn accelerometry: Validity of an open-source step-count algorithm. Sensors (Basel) 2022;22 doi: 10.3390/s22249984. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 73.Middelweerd A., van der Ploeg H.P., van Halteren A., Twisk J.W.R., Brug J., Te Velde S.J. A validation study of the Fitbit one in daily life using different time intervals. Med Sci Sports Exerc. 2017;49:1270–1279. doi: 10.1249/MSS.0000000000001225. [DOI] [PubMed] [Google Scholar]
  • 74.McClain J.J., Hart T.L., Getz R.S., Tudor-Locke C. Convergent validity of 3 low-cost motion sensors with the ActiGraph accelerometer. J Phys Act Health. 2010;7:662–670. doi: 10.1123/jpah.7.5.662. [DOI] [PubMed] [Google Scholar]
  • 75.Mikkelsen M.K., Berg-Beckhoff G., Frederiksen P., et al. Estimating physical activity and sedentary behaviour in a free-living environment: A comparative study between Fitbit Charge 2 and ActiGraph GT3X. PLoS One. 2020;15 doi: 10.1371/journal.pone.0234426. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 76.Nakagata T., Yamada Y., Taniguchi M., et al. Comparison of step-count outcomes across seven different activity trackers: A free-living experiment with young and older adults. BMC Sports Sci Med Rehabil. 2024;16:156. doi: 10.1186/s13102-024-00943-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 77.Nakagata T., Murakami H., Kawakami R., et al. Step-count outcomes of 13 different activity trackers: Results from laboratory and free-living experiments. Gait Posture. 2022;98:24–33. doi: 10.1016/j.gaitpost.2022.08.004. [DOI] [PubMed] [Google Scholar]
  • 78.Park S., Marcotte R.T., Toth L.P., et al. Free-living validation and harmonization of 10 wearable step count monitors. Transl J Am Coll Sports Med. 2021;6 doi: 10.1249/TJX.0000000000000172. [DOI] [Google Scholar]
  • 79.Pomeroy J., Brage S., Curtis J.M., Swan P.D., Knowler W.C., Franks P.W. Between-monitor differences in step counts are related to body size: Implications for objective physical activity measurement. PLoS One. 2011;6 doi: 10.1371/journal.pone.0018942. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 80.Reid R.E.R., Insogna J.A., Carver T.E., et al. Validity and reliability of Fitbit activity monitors compared to ActiGraph GT3X+ with female adults in a free-living environment. J Sci Med Sport. 2017;20:578–582. doi: 10.1016/j.jsams.2016.10.015. [DOI] [PubMed] [Google Scholar]
  • 81.Schoenfelder A., Metcalf B., Langford J., Stathi A., Western M.J., Hillsdon M. The analytical and clinical validity of the pfSTEP digital biomarker of the susceptibility/risk of declining physical function in community-dwelling older adults. Sensors (Basel) 2023;23 doi: 10.3390/s23115122. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 82.Šimůnek A., Dygrýn J., Gába A., Jakubec L., Stelzer J., Chmelík F. Validity of Garmin Vívofit and Polar Loop for measuring daily step counts in free-living conditions in adults. Acta Gymnica. 2016;46:129–135. doi: 10.5507/ag.2016.014. [DOI] [Google Scholar]
  • 83.Shimura H., Okada S., Maruo K., et al. Comparison of Omron and ActiGraph monitors in estimating daily step counts and time spent in moderate to vigorous physical activity in free-living adults. BMJ Open Sport Exerc Med. 2025;11 doi: 10.1136/bmjsem-2024-002402. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 84.Spartano N.L., Zhang Y., Liu C., et al. Agreement between Apple Watch and Actical step counts in a community setting: Cross-sectional investigation from the Framingham Heart Study. JMIR Biomed Eng. 2024;9 doi: 10.2196/54631. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 85.Sushames A., Edwards A., Thompson F., McDermott R., Gebel K. Validity and reliability of Fitbit Flex for step count, moderate to vigorous physical activity and activity energy expenditure. PLoS One. 2016;11 doi: 10.1371/journal.pone.0161224. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 86.Tedesco S., Sica M., Ancillao A., Timmons S., Barton J., O’Flynn B. Validity evaluation of the Fitbit Charge 2 and the Garmin Vivosmart HR+ in free-living environments in an older adult cohort. JMIR Mhealth Uhealth. 2019;7 doi: 10.2196/13084. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 87.Tudor-Locke C., Sisson S.B., Lee S.M., Craig C.L., Plotnikoff R.C., Bauman A. Evaluation of quality of commercial pedometers. Can J Public Health. 2006;97(Suppl. 1):S10–S16. doi: 10.1007/BF03405359. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 88.Tudor-Locke C., Ainsworth B.E., Thompson R.W., Matthews C.E. Comparison of pedometer and accelerometer measures of free-living physical activity. Med Sci Sports Exerc. 2002;34:2045–2051. doi: 10.1097/00005768-200212000-00027. [DOI] [PubMed] [Google Scholar]
  • 89.Tudor-Locke C., Barreira T.V., Schuna J.M., Jr Comparison of step outputs for waist and wrist accelerometer attachment sites. Med Sci Sports Exerc. 2015;47:839–842. doi: 10.1249/MSS.0000000000000476. [DOI] [PubMed] [Google Scholar]
  • 90.Tully M.A., McBride C., Heron L., Hunter R.F. The validation of Fitbit Zip physical activity monitor as a measure of free-living physical activity. BMC Res Notes. 2014;7 doi: 10.1186/1756-0500-7-952. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 91.Tyo B.M., Fitzhugh E.C., Bassett D.R., Jr, John D., Feito Y., Thompson D.L. Effects of body mass index and step rate on pedometer error in a free-living environment. Med Sci Sports Exerc. 2011;43:350–356. doi: 10.1249/MSS.0b013e3181e9b133. [DOI] [PubMed] [Google Scholar]
  • 92.Waddell A., Birkett S., Broom D., McGregor G., Harwood A.E. Validating the Fitbit Charge 4 wearable activity monitor for use in physical activity interventions. J Sci Med Sport. 2024;27:314–318. doi: 10.1016/j.jsams.2024.01.007. [DOI] [PubMed] [Google Scholar]
  • 93.Yano S., Koohsari M.J., Shibata A., et al. Comparison of older and newer generation Active Style Pro accelerometers in physical activity and sedentary behavior surveillance under a free-living environment. Int J Environ Res Public Health. 2019;16 doi: 10.3390/ijerph16091597. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 94.Hergenroeder A.L., Barone Gibbs B., Kotlarczyk M.P., Perera S., Kowalsky R.J., Brach J.S. Accuracy and acceptability of commercial-grade physical activity monitors in older adults. J Aging Phys Act. 2019;27:222–229. doi: 10.1123/japa.2018-0036. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 95.Ferguson T., Olds T., Curtis R., et al. Effectiveness of wearable activity trackers to increase physical activity and improve health: A systematic review of systematic reviews and meta-analyses. Lancet Digit Health. 2022;4:e615–e626. doi: 10.1016/S2589-7500(22)00111-X. [DOI] [PubMed] [Google Scholar]
  • 96.Straiton N., Alharbi M., Bauman A., et al. The validity and reliability of consumer-grade activity trackers in older, community-dwelling adults: A systematic review. Maturitas. 2018;112:85–93. doi: 10.1016/j.maturitas.2018.03.016. [DOI] [PubMed] [Google Scholar]
  • 97.Evenson K.R., Spade C.L. Review of validity and reliability of Garmin activity trackers. J Meas Phys Behav. 2020;3:170–185. doi: 10.1123/jmpb.2019-0035. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 98.Evenson K.R., Goto M.M., Furberg R.D. Systematic review of the validity and reliability of consumer-wearable activity trackers. Int J Behav Nutr Phys Act. 2015;12:159. doi: 10.1186/s12966-015-0314-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 99.Bassett D.R., Toth L.P., LaMunion S.R., Crouter S.E. Step counting: A review of measurement considerations and health-related applications. Sports Med. 2017;47:1303–1315. doi: 10.1007/s40279-016-0663-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 100.John D., Morton A., Arguello D., Lyden K., Bassett D. What is a step? Differences in how a step is detected among three popular activity monitors that have impacted physical activity research. Sensors (Basel) 2018;18 doi: 10.3390/s18041206. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 101.Ardestani M.M., Ferrigno C., Moazen M., Wimmer M.A. From normal to fast walking: Impact of cadence and stride length on lower extremity joint moments. Gait Posture. 2016;46:118–125. doi: 10.1016/j.gaitpost.2016.02.005. [DOI] [PubMed] [Google Scholar]
  • 102.Bassett D.R., John D. Use of pedometers and accelerometers in clinical populations: Validity and reliability issues. Phys Ther Rev. 2010;15:135–142. doi: 10.1179/1743288X10Y.0000000004. [DOI] [Google Scholar]
  • 103.Toth L.P., Park S., Pittman W.L., et al. Effects of brief intermittent walking bouts on step count accuracy of wearable devices. J Meas Phys Behav. 2019;2:13–21. doi: 10.1123/jmpb.2018-0050. [DOI] [Google Scholar]
  • 104.Webber S.C., St John P.D. Comparison of ActiGraph GT3X+ and StepWatch step count accuracy in geriatric rehabilitation patients. J Aging Phys Act. 2016;24:451–458. doi: 10.1123/japa.2015-0234. [DOI] [PubMed] [Google Scholar]
  • 105.Lim S.E.R., Ibrahim K., Sayer A.A., Roberts H.C. Assessment of physical activity of hospitalised older adults: A systematic review. J Nutr Health Aging. 2018;22:377–386. doi: 10.1007/s12603-017-0931-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 106.Edwardson C.L., Winkler E.A.H., Bodicoat D.H., et al. Considerations when using the activPAL monitor in field-based research with adult populations. J Sport Health Sci. 2017;6:162–178. doi: 10.1016/j.jshs.2016.02.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 107.Small S.R., Chan S., Walmsley R., et al. Self-supervised machine learning to characterize step counts from wrist-worn accelerometers in the UK Biobank. Med Sci Sports Exerc. 2024;56:1945–1953. doi: 10.1249/MSS.0000000000003478. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 108.Karas M., Strączkiewicz M., Fadel W., Harezlak J., Crainiceanu C.M., Urbanek J.K. Adaptive empirical pattern transformation (ADEPT) with application to walking stride segmentation. Biostatistics. 2021;22:331–347. doi: 10.1093/biostatistics/kxz033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 109.Rowlands A.V., Maylor B., Dawkins N.P., et al. Stepping up with GGIR: Validity of step cadence derived from wrist-worn research-grade accelerometers using the Verisense step count algorithm. J Sports Sci. 2022;40:2182–2190. doi: 10.1080/02640414.2022.2147134. [DOI] [PubMed] [Google Scholar]
  • 110.Neishabouri A., Nguyen J., Samuelsson J., et al. Quantification of acceleration as activity counts in ActiGraph wearable. Sci Rep. 2022;12 doi: 10.1038/s41598-022-16003-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 111.Koffman L., Crainiceanu C., Muschelli J. Comparing step counting algorithms for high-resolution wrist accelerometry data in NHANES 2011-2014. Med Sci Sports Exerc. 2025;57:746–755. doi: 10.1249/MSS.0000000000003616. [DOI] [PubMed] [Google Scholar]
  • 112.Gao S., Zhou X., Koffman L., et al. Comparing step-counting algorithms for high-resolution wrist accelerometry data in older adults in the ARIC study. J Gerontol A Biol Sci Med Sci. 2025;80 doi: 10.1093/gerona/glaf034. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 113.Nelson R.K., Hasanaj K., Connolly G., et al. Comparison of wrist- and hip-worn activity monitors when meeting step guidelines. Prev Chronic Dis. 2022;19:E18. doi: 10.5888/pcd19.210343. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 114.Saint-Maurice P.F., Troiano R.P., Bassett D.R., Jr, et al. Association of daily step count and step intensity with mortality among US adults. JAMA. 2020;323:1151. doi: 10.1001/jama.2020.1382. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 115.Master H., Annis J., Huang S., et al. Association of step counts over time with the risk of chronic disease in the All of Us Research Program. Nat Med. 2022;28:2301–2308. doi: 10.1038/s41591-022-02012-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 116.Hansen B.H., Dalene K.E., Ekelund U., et al. Step by step: Association of device-measured daily steps with all-cause mortality-a prospective cohort study. Scand J Med Sci Sports. 2020;30:1705–1711. doi: 10.1111/sms.13726. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 117.Consumer Technology Association Health and Fitness Technology Subcommittee . Consumer Technology Association; Arlington, VA: 2016. Physical Activity Monitoring for Fitness Wearables: Step Counting. [Google Scholar]
  • 118.Keadle S.K., Lyden K.A., Strath S.J., Staudenmayer J.W., Freedson P.S. A framework to evaluate devices that assess physical behavior. Exerc Sport Sci Rev. 2019;47:206–214. doi: 10.1249/JES.0000000000000206. [DOI] [PubMed] [Google Scholar]
  • 119.Kiefer C., Sturtz S., Bender R. Indirect comparisons and network meta-analyses. Dtsch Arztebl Int. 2015;112:803–808. doi: 10.3238/arztebl.2015.0803. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 120.Menai M., van Hees V.T., Elbaz A., Kivimaki M., Singh-Manoux A., Sabia S. Accelerometer assessed moderate-to-vigorous physical activity and successful ageing: Results from the Whitehall II study. Sci Rep. 2017;7 doi: 10.1038/srep45772. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 121.Urbanek J.K., Zipunnikov V., Harris T., Crainiceanu C., Harezlak J., Glynn N.W. Validation of gait characteristics extracted from raw accelerometry during walking against measures of physical function, mobility, fatigability, and fitness. J Gerontol A Biol Sci Med Sci. 2018;73:676–681. doi: 10.1093/gerona/glx174. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 122.Alinia P., Cain C., Fallahzadeh R., Shahrokni A., Cook D., Ghasemzadeh H. How accurate is your activity tracker? A comparative study of step counts in low-intensity physical activities. JMIR Mhealth Uhealth. 2017;5 doi: 10.2196/mhealth.6321. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 123.Pan J., Wei S. Accuracy and reliability of accelerometer-based pedometers in step counts during walking, running, and stair climbing in different locations of attachment. Sci Rep. 2024;14 doi: 10.1038/s41598-024-78684-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 124.Welk G.J., Bai Y., Lee J.M., Godino J., Saint-Maurice P.F., Carr L. Standardizing analytic methods and reporting in activity monitor validation studies. Med Sci Sports Exerc. 2019;51:1767–1780. doi: 10.1249/MSS.0000000000001966. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 125.Giurgiu M., Timm I., Becker M., et al. Quality evaluation of free-living validation studies for the assessment of 24-hour physical behavior in adults via wearables: Systematic review. JMIR Mhealth Uhealth. 2022;10 doi: 10.2196/36377. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

mmc1.pdf (6.1MB, pdf)
Download video file (6.1MB, mp4)

Articles from Journal of Sport and Health Science are provided here courtesy of Shanghai University of Sport

RESOURCES